Predicting Kidney Discard Using Machine Learning.
Predicting Kidney Discard Using Machine Learning.
复制标题
基于机器学习的肾功能预测。
DOI：
10.1097/tp.0000000000003620
复制
发表时间：
2021-09-01
期刊：
影响因子：
6.2
通讯作者：
Mehrotra S
中科院分区：
文献类型：
--
作者：
Barah M;Mehrotra S
尽管肾脏供应短缺，但在美国，每年有18%-20%的死亡供体肾脏被丢弃。2018年，有3，569个肾脏被丢弃。我们比较了机器学习（ML）技术，以识别在匹配运行时以及活检和机器灌注结果可用后有丢弃风险的肾脏。该队列由2014年12月4日至2019年7月1日期间捐献的成人死亡供体肾脏组成。研究的ML模型包括随机森林（RF），自适应提升（AdaBoost）等，并与逻辑回归（LR）进行比较。RF优于其他ML模型。在测试数据集中的8，036个废弃肾脏中，LR正确分类了3，422个肾脏，而RF正确分类了4，762个肾脏（AUC：0.85 vs 0.888，平衡准确度：0.681 vs 0.759）。在KDPI > 85%的肾脏（总共6，079个）中，RF在分类丢弃和移植预测方面显著优于LR（AUC：0.814 vs 0.717，平衡准确度：0.732 vs 0.657）。使用RF对超过388个肾脏进行了正确分类。包括活检和机器灌注变量改善了LR和RF的性能（LR的AUC：0.888和平衡准确度：0.74 vs RF的AUC：0.904和平衡准确度：0.775）。可以使用ML技术例如RF更准确地识别处于丢弃风险的肾脏。
Despite kidney supply shortage, 18%−20% deceased donor kidneys are discarded annually in the US. In 2018, 3,569 kidneys were discarded. We compared Machine Learning (ML) techniques to identify kidneys at risk of discard at the time of match-run, and after biopsy and machine perfusion results become available. The cohort consisted of adult deceased donor kidneys donated between 2014–12-04 and 2019–07-01. The studied ML models included Random Forests (RF), Adaptive Boosting (AdaBoost), among others and compared with Logistic Regression (LR). RF outperformed other ML models. Of 8,036 discarded kidneys in the test dataset, LR correctly classified 3,422 kidneys, whereas RF correctly classified 4,762 kidneys (AUC: 0.85 vs 0.888, and balanced accuracy: 0.681 vs 0.759). On the kidneys with KDPI > 85% (6,079 total), RF significantly outperformed LR in classifying discard and transplant prediction (AUC: 0.814 vs 0.717, and balanced accuracy: 0.732 vs 0.657). More than 388 kidneys were correctly classified using RF. Including biopsy and machine perfusion variables improved the performance of LR and RF (LR’s AUC: 0.888 and balanced accuracy: 0.74 vs RF’s AUC: 0.904 and balanced accuracy: 0.775). kidneys that are at risk of discard can be more accurately identified using ML techniques such as RF.